Kraballa · June 6, 2026 15:20 · Kraballa · Jun 6, 2026
diff --git a/ocrtown.py b/ocrtown.py
 # OCR Town - A simple OCR application using MeikiOCR and Tkinter
 # ----------
 # LICENSE: MIT
 # ----------
 # HOW TO USE
 # ----------
 # 0. put script in a location where you're ok with it generating 2 html files
 # 1. run script with `python ocrtown.py` 
 # 2. gui opens, select program with the selector
 # 3. make sure the program is fully visible, then click clip
 # 4. the recognised text is written to an html file. I recommend opening it with a live server so it updates in browser automatically
 # 5. use yomitan/whatever to process the text further 
 # ----------
 # LIBRARIES
 # ----------
 # the program uses the libraries below. explanation
 # 1. cv2 (OpenCV), Numpy and PIL for screenshot-taking and image processing
 # 2. the excellent meikiOCR for optical character recognition
 # 3. pywinctl (find windows), tkinter for the gui
 # 4. jinja2 to render html

 import cv2
 import numpy as np
 from meikiocr import MeikiOCR
 import tkinter as tk
 import pywinctl as pwc
 from PIL import ImageGrab
 import os
 from jinja2 import Environment, FileSystemLoader

 ocr = MeikiOCR()

 if not os.path.exists('template.html'):
    with open('template.html', 'w') as f:
        f.write("""
                <!DOCTYPE html>
                <html>

                <head>
                    <meta charset="UTF-8">
                    <title>OCR Result</title>

                    <style>
                        body {
                            font-family: "Segoe UI", sans-serif;
                            margin: auto;
                            max-width: 400px;
                            padding: 5px;
                        }
                    </style>
                </head>

                <body>
                    <h3>OCR Result</h3>
                {% for line in lines %}
                    <p>{{ line }}</p>
                {% endfor %}
                </body>
                </html>
                """)



 env = Environment(loader=FileSystemLoader('.'))
 template = env.get_template('template.html')




 if __name__ == "__main__":
    root = tk.Tk()
    root.title("OCR Town")
    root.geometry("400x300")
    label = tk.Label(root, text="OCR Town", font=("Arial", 24))
    label.pack(pady=10)

    # tkinter add dropdown with all active programs
    programs = [w.title for w in pwc.getAllWindows()]
    programs = [t for t in programs if t]
    selected_program = tk.StringVar()
    dropdown = tk.OptionMenu(root, selected_program, *programs)
    dropdown.pack(pady=10)

    def on_button_click():
        window = pwc.getWindowsWithTitle(selected_program.get())[0]
        results = []

        if(window):
            print("taking screenshot of window ", selected_program.get())
            # get window screenshot
            rect = window.getClientFrame()
            img = ImageGrab.grab(rect)
            if(not img):
                print("failed to take screenshot of window ", selected_program.get())
                return

            img = cv2.cvtColor(np.array(img), cv2.COLOR_RGB2BGR)
            output = ocr.run_ocr(img)
            results = [line['text'] for line in output if line['text']]
            print('\n'.join(results))
        else:
            print("no window found with title ", selected_program.get())

        # render template

        rendered = template.render(lines=results)
        # save rendered html to file
        with open("output.html", "wb") as f:
            f.write(rendered.encode('utf-8'))
        pass

    button = tk.Button(root, text="Clip", command=on_button_click, font=("Arial", 36))
    button.pack(pady=10)
    root.mainloop()
	# OCR Town - A simple OCR application using MeikiOCR and Tkinter
	# ----------
	# LICENSE: MIT
	# ----------
	# HOW TO USE
	# ----------
	# 0. put script in a location where you're ok with it generating 2 html files
	# 1. run script with `python ocrtown.py`
	# 2. gui opens, select program with the selector
	# 3. make sure the program is fully visible, then click clip
	# 4. the recognised text is written to an html file. I recommend opening it with a live server so it updates in browser automatically
	# 5. use yomitan/whatever to process the text further
	# ----------
	# LIBRARIES
	# ----------
	# the program uses the libraries below. explanation
	# 1. cv2 (OpenCV), Numpy and PIL for screenshot-taking and image processing
	# 2. the excellent meikiOCR for optical character recognition
	# 3. pywinctl (find windows), tkinter for the gui
	# 4. jinja2 to render html

	import cv2
	import numpy as np
	from meikiocr import MeikiOCR
	import tkinter as tk
	import pywinctl as pwc
	from PIL import ImageGrab
	import os
	from jinja2 import Environment, FileSystemLoader

	ocr = MeikiOCR()

	if not os.path.exists('template.html'):
	with open('template.html', 'w') as f:
	f.write("""
	<!DOCTYPE html>
	<html>

	<head>
	<meta charset="UTF-8">
	<title>OCR Result</title>

	<style>
	body {
	font-family: "Segoe UI", sans-serif;
	margin: auto;
	max-width: 400px;
	padding: 5px;
	}
	</style>
	</head>

	<body>
	<h3>OCR Result</h3>
	{% for line in lines %}
	<p>{{ line }}</p>
	{% endfor %}
	</body>
	</html>
	""")



	env = Environment(loader=FileSystemLoader('.'))
	template = env.get_template('template.html')




	if __name__ == "__main__":
	root = tk.Tk()
	root.title("OCR Town")
	root.geometry("400x300")
	label = tk.Label(root, text="OCR Town", font=("Arial", 24))
	label.pack(pady=10)

	# tkinter add dropdown with all active programs
	programs = [w.title for w in pwc.getAllWindows()]
	programs = [t for t in programs if t]
	selected_program = tk.StringVar()
	dropdown = tk.OptionMenu(root, selected_program, *programs)
	dropdown.pack(pady=10)

	def on_button_click():
	window = pwc.getWindowsWithTitle(selected_program.get())[0]
	results = []

	if(window):
	print("taking screenshot of window ", selected_program.get())
	# get window screenshot
	rect = window.getClientFrame()
	img = ImageGrab.grab(rect)
	if(not img):
	print("failed to take screenshot of window ", selected_program.get())
	return

	img = cv2.cvtColor(np.array(img), cv2.COLOR_RGB2BGR)
	output = ocr.run_ocr(img)
	results = [line['text'] for line in output if line['text']]
	print('\n'.join(results))
	else:
	print("no window found with title ", selected_program.get())

	# render template

	rendered = template.render(lines=results)
	# save rendered html to file
	with open("output.html", "wb") as f:
	f.write(rendered.encode('utf-8'))
	pass

	button = tk.Button(root, text="Clip", command=on_button_click, font=("Arial", 36))
	button.pack(pady=10)
	root.mainloop()
No results found